Warp <war### [at] tagpovrayorg> wrote...
>   Ok, the pass_through helped.
>   It's just not very intuitive. If the light passes through transparent
> objects, why not the photons? I think many people will be confused by this
> feature.

Yes, I agree that it is kind of confusing.  Unfortunately, there's not an
easy way around it.  The problem is that, for a photon to be 'valid', it has
to hit the 'target' object.  If it doesn't hit the target object, it is
discarded (for energy conservation reasons).  Sometimes, though, you will
put the target behind a clear object, such as a media container.  When this
happens, you have to put pass_through in the object that blocks the path
from the light to the target.  Then, MegaPov will ignore the pass_through
object and let the photon hit the target.

>   I made the container box a lot smaller and it looks a lot better now.
> The reflected beams are denser and the sharp line is much less obvious
now.

Looks great!  The errors were probably estimation errors caused by too few
samples.  The smaller container box means that the samples would be packed
closer together and therefore would produce a better estimation.

>   The second image uses method 2. There's a strange non-lit line where
there
> was a sharp edge before (at the beginning of the first reflected beam).
Also
> the light reflected from the lens is almost gone.

When you use method 2, you'll have to increase the number of samples.
Otherwise, some of your samples will miss important parts of the scene.  For
example, you probably had samples hit before and after the reflected caustic
from the lens, but no samples hit directly on it.  Adding some jitter might
help a little bit, but nothing works as well as additional samples.

-Nathan

Post a reply to this message